Search Advance

Hybrid semantic + keyword search across every doc, product and record — ranked by meaning, answered with sources.

Search Advance is a Next.js + pgvector search layer built for MADXVISION's own stack and client platforms. Content is chunked and embedded on write, then queried with a hybrid ranker that blends keyword match, vector similarity and freshness. Every result carries its source record so an answer traces back to the exact document, product page or database row it came from. Sync workers keep the index current as content changes, and an admin console tracks crawl status, index size and slow queries.

• Hybrid keyword + vector ranking on every query
• Source-linked answers — trace back to the exact record
• Incremental re-index on content change, no full rebuild
• Typo-tolerant, multilingual query matching
• Admin console for crawl status, index size and slow queries
• Drop-in search bar component for any MADX site

• Next.js
• pgvector
• Hybrid rank
• Embeddings
• Source citations
• Admin console